Instructions to use YanweiLi/MGM-34B-HD with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use YanweiLi/MGM-34B-HD with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="YanweiLi/MGM-34B-HD")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("YanweiLi/MGM-34B-HD", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use YanweiLi/MGM-34B-HD with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "YanweiLi/MGM-34B-HD"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "YanweiLi/MGM-34B-HD",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/YanweiLi/MGM-34B-HD

SGLang

How to use YanweiLi/MGM-34B-HD with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "YanweiLi/MGM-34B-HD" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "YanweiLi/MGM-34B-HD",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "YanweiLi/MGM-34B-HD" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "YanweiLi/MGM-34B-HD",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use YanweiLi/MGM-34B-HD with Docker Model Runner:
```
docker model run hf.co/YanweiLi/MGM-34B-HD
```

MGM-34B-HD / README.md

YanweiLi

Use MGM to represent Mini-Gemini

eddd6e0 verified about 2 years ago

preview code

raw

history blame contribute delete

2.31 kB

	---
	tags:
	- vision-language model
	- llama
	- yi
	- generation
	datasets:
	- YanweiLi/MGM-Instruction
	---

	# MGM-34B-HD Model Card
	<a href='https://github.com/dvlab-research/MGM'><img src='https://img.shields.io/badge/Project-Code-violet'></a>
	<a href='https://mini-gemini.github.io/'><img src='https://img.shields.io/badge/Project-Page-Green'></a>
	<a href='https://arxiv.org/pdf/2403.18814.pdf'><img src='https://img.shields.io/badge/Paper-Arxiv-red'></a>

	## Model details
	The framework supports a series of dense and MoE Large Language Models (LLMs) from 2B to 34B with HD image understanding, reasoning, and generation simultaneously.
	You can also try our other MGM series models:

	Normal resolution setting: [MGM-2B](https://huggingface.co/YanweiLi/MGM-2B), [MGM-7B](https://huggingface.co/YanweiLi/MGM-7B), [MGM-13B](https://huggingface.co/YanweiLi/MGM-13B), [MGM-8x7B](https://huggingface.co/YanweiLi/MGM-8x7B), [MGM-34B](https://huggingface.co/YanweiLi/MGM-34B)

	High resolution setting: [MGM-7B-HD](https://huggingface.co/YanweiLi/MGM-7B-HD), [MGM-13B-HD](https://huggingface.co/YanweiLi/MGM-13B-HD), [MGM-8x7B-HD](https://huggingface.co/YanweiLi/MGM-8x7B-HD)

	Model type:
	MGM is an open-source chatbot trained by fine-tuning Nous-Hermes-2-Yi-34B on GPT-generated multimodal instruction-following data.

	It empowers existing frameworks to support HD image understanding, reasoning, and generation simultaneously.

	Model version:
	MGM HD Version with LLM Nous-Hermes-2-Yi-34B

	Model date:
	MGM-34B-HD was trained on 03/2024.

	## License
	Nous-Hermes-2-Yi-34B is licensed under the apache-2.0 License,

	Where to send questions or comments about the model:
	https://github.com/dvlab-research/MGM/issues

	## Intended use
	Primary intended uses:
	The primary use is research on large multimodal models and chatbots.

	Primary intended users:
	The primary intended users of the model are researchers and hobbyists in computer vision, natural language processing, machine learning, and artificial intelligence.

	## Training data
	This model is trained based on [MGM-Instruction](https://huggingface.co/datasets/YanweiLi/MGM-Instruction) dataset, please to the [Github](https://github.com/dvlab-research/MGM) for more detail.

	## Acknowledgement
	This project is not affiliated with Google LLC.